The present disclosure generally relates to image processing, and, in particular, to processing a satellite image of a region to determine a terrain boundary included therein.
Satellite imagery consists of photographs of the earth, and can be acquired at a variety of different resolutions from a variety of different sensors. Conventional systems provide for processing the satellite imagery to distinguish between different types of terrains (e.g., bodies of water, land, mountains, forests). Such processing can be useful, for example, in web mapping service applications.
Due to the above-described variance in resolution and sensor capture for the satellite imagery, it may be difficult to automatically refine maps based on captured satellite image data. For example, some image data may have low resolution, and the processing of such data to determine terrain boundaries may yield inaccurate results (e.g., due to noise in the image). Thus, it may be desirable to refine the terrain boundaries of maps corresponding to multiple satellite images.